The dilemma and countermeasures of educational data ethics in the age of intelligence

The integration of artificial intelligence (AI), learning analytics, data mining, cloud computing, and other technologies into education has created an era of intelligent education. While these technologies offer significant potential for improving teaching and learning, they also raise critical ethical concerns related to the collection, management, sharing, and use of educational data. This paper investigates the ethical dilemmas surrounding educational data in the context of intelligent education. International organizations, such as the United Nations and UNESCO, have recognized the importance of data ethics and have issued various recommendations for data privacy and security. The 2021 EDUCAUSE Horizon Report highlighted several key technologies and practices that affect higher education information security, including cloud vendor management, endpoint detection and response, multifactor authentication, preserving data authenticity, research security, and student data privacy. These reports and initiatives underscore the growing global concern about data ethics in education. The study aims to define educational data ethics, reviewing various definitions from organizations like the Center for Democracy & Technology (CDT), the General Services Administration (GSA), and the Open Data Institute (ODI). While these definitions offer different perspectives, they all emphasize the importance of responsible data handling to avoid harm to individuals and the public interest. The study further examines national frameworks and legal instruments from various countries, such as the U.S. General Services Administration's Data Ethics Framework, the UK's Government Digital Service Data Ethics Framework, and China's Data Security Law, Personal Information Protection Law, and Internet Data Security Management Regulations. This context establishes the necessity for a systematic review of the international research on educational data ethics to develop specific and systematic solutions applicable to the Chinese context, particularly addressing the vulnerabilities of learners, especially children.

The study employed a bibliometric analysis of literature from the Web of Science database, focusing on keywords related to "data," "education," and "ethics." Using ASReview software, 385 papers were selected from 88,764 initial results. CiteSpace (6.1.R4) was used for visualization and analysis of research hotspots, trends, and key scholars and institutions. The analysis identified key research areas, including learning analytics, data science, artificial intelligence, big data, gender bias, and the importance of artificial intelligence literacy. The analysis of research trends showed three stages: (1) 2019-2020 focused on information extraction and big data analytics; (2) 2020-2021 emphasized privacy principles; and (3) 2021-2023 highlighted systematic reviews, user acceptance, educational data mining, teacher roles, and decision-making processes. Analysis of authors and institutions revealed key contributors and their areas of focus within the field.

The research utilized a mixed-methods approach combining bibliometric analysis and in-depth qualitative analysis. The bibliometric analysis involved a systematic search of the Web of Science database using keywords related to "data," "education," and "ethics." The search yielded 88,764 results, which were then filtered using ASReview software, a machine learning-based tool for efficient literature screening. After filtering based on predefined criteria (e.g., relevance to educational data ethics, theoretical or practical contribution, focus on technology solutions or case studies), a final set of 385 papers were included in the analysis. CiteSpace software was then employed to conduct a bibliometric analysis of the selected papers to identify research hotspots, trends, and key contributors. This involved analyzing keyword co-occurrence networks, timeline views, and author and institution co-citation networks. The results of the bibliometric analysis guided the selection of papers for in-depth qualitative review. A detailed qualitative analysis involved a comprehensive review of key papers to identify recurring themes and patterns in the literature, particularly focusing on the common dilemmas faced by researchers in the field of educational data ethics and the proposed countermeasures and solutions. This mixed-methods approach allowed for a comprehensive understanding of the landscape of research on educational data ethics, providing a robust basis for drawing conclusions and making recommendations.

The study identified three primary dilemmas in educational data ethics: (1) Privacy violations during data collection, storage, and sharing. This includes issues with informed consent, anonymization of sensitive data, and potential leakage during data sharing. (2) The deprivation of independent choice due to the predictive function of educational data. Over-reliance on predictions can limit learners' opportunities for trial and error, hinder innovative thinking, and create information cocoons. For educators, it can lead to overdependence on data, reducing their critical thinking and professional development. (3) Data-driven evaluation lacking a "forgetting ability." The permanent storage of data can solidify labels and hinder learners' development, while also creating a "chilling effect" that restricts learners' behavior and limits opportunities for improvement. The study further revealed that existing research on educational data ethics often focuses on specific technologies or contexts. While there have been horizontal solutions proposed, they lack systematization and fail to fully address the multifaceted nature of the problem. The bibliometric analysis highlighted the increasing importance of learning analytics and artificial intelligence in educational research, which have emerged as significant areas of concern regarding data ethics. The analysis also identified leading scholars and institutions, providing insight into the network and directions of current research.

The findings of this study demonstrate the significant ethical challenges associated with the use of educational data in the context of intelligent education. The three key dilemmas identified—privacy violations, the restriction of independent choice, and the absence of a "forgetting ability"—highlight the need for a learner-centered approach to educational data ethics. Existing solutions often lack a comprehensive, systematic framework, leading to inconsistent and ineffective strategies. The study's bibliometric analysis further corroborates these findings, showcasing the evolving research landscape and the key areas of focus. By integrating technological solutions like blockchain, 5G technology, and federated learning, along with systematic standards and ethical education, the study proposes a more comprehensive framework to address the identified dilemmas. This integrated strategy fosters responsible data use, protects learner privacy, and promotes autonomy while maintaining the beneficial aspects of intelligent education technologies.

This study provides a systematic analysis of the dilemmas and countermeasures of educational data ethics in the age of intelligent education. The identification of three key challenges—privacy violation, restriction of independent choice, and the lack of "forgetting ability"—along with the proposed learner-centered strategies, offers valuable insight for future research and practice. Future research should prioritize practical implementation of these strategies, focusing on developing and testing effective mechanisms for data governance, privacy protection, and ethical education. This includes further investigation into the application of technologies like blockchain and federated learning within specific educational contexts.

The study primarily relies on a bibliometric analysis and in-depth literature review, lacking direct empirical evidence from specific research practices. While the bibliometric analysis provides a valuable overview of research trends, it does not capture the nuances of practical implementations in various educational settings. The proposed strategies are largely theoretical and require further empirical testing to assess their effectiveness in different contexts. Future research should focus on rigorous empirical studies to validate the effectiveness of these strategies and address the limitations of the present study.